The TEQSA generative AI framework is not a single document and it is not a standard. It is a set of resources published between 2023 and 2025, together with the June 2024 request for information, that tell providers how TEQSA now reads five existing Threshold Standards: 1.4 on assessment, 5.2 on academic integrity, 3.2 on staffing, 6.3 on academic governance and 7.1 on information for students. A compliance officer who maps each resource to the standard it serves can build the evidence trail an assessor will ask for.
This article does that mapping. It draws on fifteen years of TEQSA work and on the AI action plans I have reviewed since 2024, most of them strong on intent and weak on evidence.
What the TEQSA generative AI framework consists of
Three publications and one regulatory action make up the framework. The 2023 paper Assessment reform for the age of artificial intelligence set out two principles: that students should be equipped for a society pervaded by AI, and that trustworthy judgments about learning require multiple, inclusive and contextualised approaches to assessment. The June 2024 request for information required every provider to lodge, by 3 July 2024, a credible institutional action plan overseen by governance.
The 2025 paper Enacting assessment reform in a time of artificial intelligence moved from principle to practice, and the gen-AI knowledge hub collects TEQSA's resources alongside sector examples. Our summary of the generative AI toolkit published by TEQSA describes each resource. None is a Threshold Standard; each is guidance on how existing standards will be applied, which is why the mapping matters.
Standard 1.4: assessment that confirms attainment
Standard 1.4 requires assessment capable of confirming that every learning outcome has been achieved. The 2023 and 2025 papers are, in substance, TEQSA's account of what that now requires: program-level design, secured assessment points where a student's own attainment is verified, and a preference for assessing process as well as product. The evidence is a course-level assessment map showing which tasks are secured and confirming each course learning outcome is evidenced through at least one of them, approved by the academic board and reviewed on a cycle.
TEQSA has been explicit that detection software is not sufficient on its own. An assessor will look for the map, for unit outlines that have changed since 2023, and for moderation records showing the redesigned tasks working. A plan to redesign assessment is not evidence that assessment has been redesigned.
Standard 5.2: integrity as a system
Standard 5.2 requires policies, training, monitoring and action to protect academic integrity. The request for information was directed at this standard as much as at 1.4, and the action plans TEQSA asked for were expected to address integrity risk institution-wide. The evidence trail is the plan itself, the board minute adopting it, the policy amendments it produced, the staff and student training records, the case register showing AI-related breaches investigated and decided, and the annual integrity report to the academic board.
The most common gap I see is the last item. Providers have the policy and the plan but cannot show that anyone has reviewed whether the plan worked. The academic integrity compliance guidance we publish sets out the reporting cycle that closes that gap.
Standard 3.2: staff who can teach and assess with AI in the room
Standard 3.2 concerns staffing, and specifically that staff are qualified, supported and developed for the teaching they do. The framework's first principle, equipping students for an AI-pervaded society, cannot be met by staff who have not themselves been equipped. Assessors ask what professional development on generative AI has been delivered, who attended, and how sessional staff were included.
The evidence is a professional development register with dates and attendance, revised marking guidance that addresses AI-generated work, and evidence that staff feedback shaped the redesign. Where development exists only as a policy commitment, the assessor will infer that the front line has not changed.
Standard 6.3: governance that owns the response
The request for information required the action plan to be overseen by governance, and Standard 6.3 places responsibility for academic quality and integrity with the academic board. TEQSA reads the minutes. It expects to see the academic board receive the plan, question it, adopt it, receive progress reports and adjust it, with the corporate board informed. A plan signed by the CEO and never tabled is a plan the governing body did not own, and the non-delegation principle applies here as it does everywhere else.
This is where the framework and the shift to self-assurance meet. A self-assurance report that asserts a governed response to generative AI must point to minutes that show it, and generic assertions drafted by the same tools the plan is meant to govern are, in my experience, read with particular care.
Standard 7.1: telling students the truth about AI
Standard 7.1 requires information provided to students to be accurate, clear and not misleading. Students need to know what use of generative AI is permitted in each task, how it must be acknowledged, and what follows a breach. Unit outlines, assessment instructions and the academic integrity policy are the evidence, and inconsistency between them is the finding. The article on what compliance officers want to see on AI use covers the student-facing material in detail.
Building the evidence trail
Put together, the TEQSA generative AI framework mapping gives a compliance officer a single index: for each standard, the resource that informs TEQSA's reading, the documents showing the provider's response, and the minute proving governance owned it. That index is short, and it is the difference between an action plan and a demonstrable one. TEQSA said in 2024 that it would follow up insufficient plans and consider its regulatory tools; the plans it will find insufficient are the ones that never became records.
Download the AI Governance Checklist
— a one-page map of TEQSA's generative AI resources to Standards 1.4, 5.2, 3.2, 6.3 and 7.1 with the evidence each requires, drawn from our TEQSA registration and governance work with private providers. Get the checklist
Want the full article?
Enter your email for free access to the rest of this guide and our TEQSA resource library.
Frequently asked questions
Is TEQSA's generative AI guidance legally binding?
No. The 2023 and 2025 papers and the knowledge hub are guidance, not Threshold Standards. They describe how TEQSA applies existing standards such as 1.4, 5.2 and 6.3, and it is those standards that are enforceable.
What did the June 2024 request for information require?
Every registered provider had to lodge, by 3 July 2024, a credible institutional action plan addressing the risk generative AI poses to award integrity, with the plan overseen by the provider's governance. TEQSA indicated it would follow up insufficient or absent plans.
Does TEQSA accept AI detection software as a control?
Not on its own. TEQSA's guidance favours program-level assurance with secured assessment points where a student's own attainment is verified, and treats detection tools as at most one element of a wider design.
Which body should oversee a provider's AI action plan?
The academic board, under Standard 6.3, with reporting to the corporate governing body. The minutes should show the plan adopted, progress reported and adjustments made.
Dr Brendan Moloney is CEO of Darlo Higher Education, Australia's largest specialist TEQSA consultancy. He holds a PhD from the University of Melbourne, is a Cambridge University Press author on governance in higher education, and has advised private providers on registration and course accreditation for more than fifteen years.